y**3
发帖数: 267 | 1 I have only 7 yearly data points for a time series. Is it possible to do a
time series analysis to predict 8 year 9yrtc?
For a Arima model, minimum 40-50 data points is required.
Is there any expert with time series can help?\
thanks |
c**d
发帖数: 104 | 2 You can use the following 4 methods to do forecasting for a short time
series.
My previous boss (MBA degree)always asked this kind of questions: to predict
next 1 or 2 points with 7 or 8 time points. Believe or not, sometime it
worked very well.
1: Moving averages
2: Forecasting using exponential smoothing
3: Accounting for data trend using Holt's smoothing
4: Accounting for data seasonality using Winter's smoothing
【在 y**3 的大作中提到】
: I have only 7 yearly data points for a time series. Is it possible to do a
: time series analysis to predict 8 year 9yrtc?
: For a Arima model, minimum 40-50 data points is required.
: Is there any expert with time series can help?\
: thanks
|
h***i
发帖数: 3844 | 3 我觉得，这种事情，肉眼看看就得了，做model没什么意思。
【在 y**3 的大作中提到】
: I have only 7 yearly data points for a time series. Is it possible to do a
: time series analysis to predict 8 year 9yrtc?
: For a Arima model, minimum 40-50 data points is required.
: Is there any expert with time series can help?\
: thanks
|
y**3
发帖数: 267 | 4 Thanks.
Does moving average require smaller data points? If so, how many |
A****1
发帖数: 33 | 5 My two cents:
When you have a time series data, first, take a look at the data. Is there
any trend or seasonality? If the trend or seasonality is very obvious (e.g.
logistic function), you can do forecasting even though there is a limited
number of data points. If the seven points look like random process, it is
hard to predict.
【在 y**3 的大作中提到】
: I have only 7 yearly data points for a time series. Is it possible to do a
: time series analysis to predict 8 year 9yrtc?
: For a Arima model, minimum 40-50 data points is required.
: Is there any expert with time series can help?\
: thanks
|
y**3
发帖数: 267 | 6 thanks.
Yes, there seems a strong linear trend. Can I just fit the reponse vs time
and then extrapolate to beyond observed time point? It seems not
statistically sound.
Or you meant still use ARMA or MA to do the forecasting?
.
【在 A****1 的大作中提到】
: My two cents:
: When you have a time series data, first, take a look at the data. Is there
: any trend or seasonality? If the trend or seasonality is very obvious (e.g.
: logistic function), you can do forecasting even though there is a limited
: number of data points. If the seven points look like random process, it is
: hard to predict.
|
P****D
发帖数: 11146 | 7 For business purpose, I don't see why not.
【在 y**3 的大作中提到】
: thanks.
: Yes, there seems a strong linear trend. Can I just fit the reponse vs time
: and then extrapolate to beyond observed time point? It seems not
: statistically sound.
: Or you meant still use ARMA or MA to do the forecasting?
:
: .
|
